Jump models switch infrequently between states to fit a sequence of data while taking the ordering of the data into account. We propose a new framework for joint feature selection, parameter and state-sequence estimation in jump models. Feature selection is necessary in high-dimensional settings...